S156 • JID 2010:201 (Suppl 2) • Gottlieb et al
S U P P L E M E N T A R T I C L E
Screening and Treatment to Prevent Sequelae
in Women with Chlamydia trachomatis Genital
Infection: How Much Do We Know?
Sami L. Gottlieb,1Stuart M. Berman,1and Nicola Low2
1Division of STD Prevention, Centers for Disease Control and Prevention, Atlanta, Georgia, and2Department of Social and Preventive Medicine, University of Bern, Switzerland
Background. An important question for chlamydia control programs is the extent to which finding and treating prevalent, asymptomatic Chlamydia trachomatis genital infection reduces reproductive sequelae in infected women.
Methods. We reviewed the literature to critically evaluate evidence on the effect of chlamydia screening on development of sequelae in infected women.
Results. Two randomized controlled trials of 1-time screening for chlamydial infection—in a Seattle-area health maintenance organization and a Danish school district—revealed that screening was associated with an∼50% reduction in the incidence of pelvic inflammatory disease over the following year. However, both of these trials had methodological issues that may have affected the magnitude of observed screening benefits and might limit generalizability to other populations. A large, nonrandomized cohort of chlamydia screening among US Army recruits, although limited by lack of outpatient data, did not find a benefit of similar magnitude to the randomized trials. Methodological limitations restrict valid conclusions about individual benefits of screening using data from historical cohorts and ecological studies. We identified no trials directly evaluating the effect of chlamydia screening on subclinical tubal inflammation or damage, ectopic pregnancy, or tubal factor infertility and no studies addressing the effects of11 round of screening, the optimal frequency of screening, or the benefits of screening for repeat
infections.
Conclusions. Additional studies of the effectiveness of chlamydia screening would be valuable; feasible study designs may depend on the degree to which screening programs are already established. In addition, better natural history data on the timing of tubal inflammation and damage after C. trachomatis infection and development of more accurate, noninvasive tools to assess chlamydial sequelae are essential to informing chlamydia control efforts. Routine screening for asymptomatic Chlamydia
tra-chomatis genital infection among young, sexually active
women is a fundamental component of chlamydia con-trol in many countries [1, 2]. Thus, a critical question for chlamydia control programs is the extent to which finding and treating prevalent, asymptomatic
chlamyd-Potential conflicts of interest: None reported. Financial support: None reported.
Supplement sponsorship: This article is part of a supplement entitled “Chlamydia
trachomatis Genital Infection: Natural History, Immunobiology, and Implications for
Control Programs,” which was sponsored by the Centers for Disease Control and Prevention.
The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention. Reprints or correspondence: Sami L. Gottlieb, Div of STD Prevention, Centers for Disease Control and Prevention, 1600 Clifton Rd NE, MS E-02, Atlanta, GA 30333 (sgottlieb@cdc.gov).
The Journal of Infectious Diseases 2010; 201(S2):S156–S167
2010 by the Infectious Diseases Society of America. All rights reserved. 0022-1899/2010/20112S2-0009$15.00
DOI: 10.1086/652396
ial infections reduces the incidence of reproductive tract sequelae in infected women. Many uncertainties remain about the natural history of untreated C. trachomatis infection in leading to sequelae, as Haggerty et al discuss in this supplement [3]. One of the most important factors affecting the potential effectiveness of chlamydia control programs is the timing of tubal inflammation and damage relative to the acquisition of infection. This timing profoundly impacts the likelihood that infec-tions can be identified and treated before development of short-term sequelae, such as symptomatic pelvic in-flammatory disease (PID), and before development of tubal damage that may ultimately lead to long-term sequelae, such as tubal factor infertility and ectopic pregnancy. There is little doubt that C. trachomatis in-fection can lead to sequelae [4] and that chlamydial infection can be effectively treated with available
Screening and Treatment to Prevent Sequelae • JID 2010:201 (Suppl 2) • S157
antibiotics [5]. What is less clear, however, is the degree to which treatment given at various times during the natural course of infection can interrupt progression to sequelae.
The duration of genital C. trachomatis infection in the ab-sence of treatment has not been completely elucidated [6], but available long-term studies suggest that a typical untreated chla-mydial infection can last a year or longer [7, 8]. Thus, asymp-tomatic C. trachomatis infections may have been present for many months when detected by screening. By assuming that women acquire new chlamydial infections at a steady rate be-tween screenings and that women are screened consistently at 1-year intervals, the mean duration of each chlamydia infection detected would be 6 months. With longer screening intervals, as frequently occurs in the United States [9], duration of in-fection could be greater. If acute PID and/or tubal damage primarily occur very soon after acquisition of C. trachomatis infection, screening will only affect that small proportion of the infected population who have recently acquired infection. Screening at less frequent intervals would result in even fewer women receiving benefit. If, however, tissue-damaging patho-logic processes are mainly initiated later in the course of in-fection or continue to be elicited by ongoing inin-fection over an extended period of time, screening would benefit a greater pro-portion of infected women.
In preparation for the April 2008 Chlamydia Immunology and Control Expert Advisory Meeting, we reviewed and criti-cally evaluated evidence for the effectiveness of chlamydia screening in reducing the incidence of sequelae in screened women. We assessed this evidence in the context of achieving better understanding of chlamydial natural history and path-ogenesis; thus, we focused primarily on the role of screening in interrupting progression to upper tract inflammation and sequelae in already infected women. We did not explore the effect of screening or other control efforts in reducing trans-mission of C. trachomatis in a population, which could prevent sequelae by preventing new incident chlamydial infections in women in the population. This evidence has recently been re-viewed elsewhere [10]. Our goals were to assess what is known about screening and treatment to prevent sequelae in already infected women, to delineate remaining gaps in knowledge that have implications for chlamydia control programs, and to out-line the most important research needs to fill those gaps.
METHODS
We conducted a search of the English-language literature from 1960 through March 2008 with use of the Medline comput-erized database of the US National Library of Medicine. The Medical Subject Headings and free text terms “chlamydia,”
“Chlamydia trachomatis,” and “chlamydia infections”
(ex-ploded) were combined with the terms “pelvic inflammatory disease,” “salpingitis,” “endometritis,” “infertility,” and
“ec-topic pregnancy” (exploded) and then limited to humans and persons⭓13 years of age. The search excluded articles focused primarily on “Chlamydophila pneumoniae.” We also searched reference lists of articles to identify potential additional references.
We included original, prospective studies of chlamydia screening interventions in women that had any of the following primary outcomes: PID, infertility, or ectopic pregnancy. We also included registry-based retrospective cohort studies and ecological studies if they discussed one of the primary outcomes in relation to chlamydia testing. We did not include cross-sectional and case-control studies, nor did we include studies with a primary outcome of screening coverage, chlamydia prev-alence or incidence, or adverse pregnancy outcomes other than ectopic pregnancy. We also excluded articles that assessed screening and outcomes after therapeutic abortion or other surgical instrumentation.
RESULTS
Our initial literature search yielded 875 unique citations. After review by an author (S.G.) for the aforementioned inclusion criteria, a total of 11 articles were selected for critical evaluation: 2 randomized controlled trials [11, 12] and 1 nonrandomized cohort study [13] that evaluated chlamydia screening inter-ventions and clinical PID outcomes; 4 registry-based retro-spective cohort studies evaluating chlamydia testing and long-term outcomes, including ectopic pregnancy [14–17] and infertility [17]; and 4 ecological studies of rates of chlamydia infection and either PID [18–20] or ectopic pregnancy out-comes (Table 1) [18, 21]. We identified no randomized or nonrandomized trials of the effect of chlamydia screening on subclinical tubal inflammation, tubal scarring, or the long-term outcomes of ectopic pregnancy and tubal factor infertility. We present the results according to study methods.
Randomized controlled trials. A randomized controlled trial by Scholes et al [12] is the study cited most widely, in-cluding by the US Preventive Services Task Force [22], as pro-viding evidence for the individual benefits of screening. In this study involving 2607 young women at high risk in a Seattle-area health maintenance organization (HMO) population, a 1-time proactive approach of inviting women for chlamydia screening (64% were tested) significantly reduced the risk of subsequent PID by∼50% over the next year, compared with a control group not invited for testing (relative risk, 0.44; 95% confidence interval, 0.20–0.88). At the time of this review, these data remain the best available on the benefits of screening to prevent PID in infected women. However, some aspects of this trial raise questions as to whether the magnitude of observed benefits is generalizable and provides realistic expectations of benefit in real-world settings.
S158 Table 1. St udies of the Indivi dual Be nefits of Chlamy d ia tra choma tis Scre ening on Pelvi c Inflamm a tory Di se ase (P ID), Ectopic Pre gnancy (EP), and Infert ility Reference (year) Stu d y de s ign and setting Population Study methods Outcome measures Reported findings Strengths Limitations O v e ral l rele va n c e a n d im port an t mes sage s RCTs Scho le s e t a l [12 ] (19 96 ) RCT of C. tr a cho m ati s screening to reduce PID; HM O in Sea tt le area (1990– 1992) Enr o lle d sin gl e , sexua lly activ e w o m e n aged 18– 34 yea rs wh o re -tu rned s u rve y a n d h a d ris k scor e ⭓ 3( 1p o in t for age ! 25 yea rs , ⭓ 2 s e x p a rtner s in 1 2 months, nulligravidity , o r d o uch ing; 2 p o int s for bla ck ra ce); 36 ,5 4 7 w o men ra n dom ize d; 5 7 % retur ned s u rve y; 260 7 (7 % of ra nd o m -ized ) h a d score ⭓ 3; 81% we re age d 1 8 – 24 years ; 22% of wo me n wer e b lack Rando m ly assig ne d to screening (1009 women inv it ed to 1-t im e C. tr a -c h o mat is test ing b y E L IS A a n d/o r c u lture) o r c o n trol (1 5 9 8 in u su a l care) O u tco m e a t 1 2 m o n ths: PID inc id e n c e , ascer -ta ined b y s e lf -rep ort, inp a ti e n t dia gn os is c o d es, p h a rmac y re-c o rds, o utpa tient code s fo r PI D and cer v ic iti s , la bo rato ry re c o rds; m e d ica l rec-ord re v iew to confi rm d ia g n o s is ; int entio n to scree n analy sis S c reen ing g rou p: 6 4 % tes ted, 7 % (44 ) had C . tr a c h omat is infe ct ion and were treated; controls: p er -c e n tag e test ed n o t repo rted; o u t-come s : 42 PI D cases; scre e n in g group: 9 cases (7 of 9 had been tested); controls: 33 cases; inci-den c e of PID : 8 cases pe r 1 0 ,0 0 0 P M in s c reenin g g roup, 18 cases per 10 ,0 0 0 PM in th e control group (RR, 0.44; 95% CI, 0.2 0– 0. 88 ; n o cha ng e in R R aft e r adju s tm e n t fo r b a sel in e cha ra c-teristics); 7 o f 4 2 w omen w ith PID ha d p o s iti v e C. tr a cho ma ti s / Nei s seri a g o n o rr h o e a e tes t re -sult s a t ti m e o f d ia g n o s is Randomized controlled des ig n; la rg e samp le size; intention to scree n analy sis Randomized before enroll-men t and scre e n in g g roup p refer entia lly calle d to sub mi t sur v e y and set up te s ti n g ; on ly 7 % o f thos e ran do m-ized enrolled, and 1 :1.6 ins te a d of pl ann e d 1 :2 screening to control ra-tio sug ge s ts ra ndo m iz a -tio n may ha ve bee n c o m prom ise d ; S e a ttle HMO, women at high ri s k ; thus, fi ndi ngs m a y not ge ne ra liz e to oth e r p o pul at ion s Stu d y mo s t fr e q u e n tl y cite d a s evi-d e nce fo r the b enefi ts o f C. tr a -c h o mat is s c reeni ng, s u g g est in g tha t a 1-ti m e in v ita ti o n fo r screening can prevent about half of subsequent PID cases; magni-tud e o f P ID pre v e n te d app ea rs h ighe r than w hat m ig h t b e e x-pec te d b a sed on na tu ra l h isto ry s tu dies ; thi s a n d iden tified lim ita-ti ons, ra is e q ues tions a s to w h ether fi ndi ngs a re g ene ral iz -able and pr o vid e rea listi c e x p e c -ta tions o f b ene fit in real -wor ld s e ttings Øste rg aard et al [1 1] (2 00 0 ) Clus te r RCT of 17 hi g h schoo ls in Den m ark (19 97 – 199 8 ), com p a r-in g in v itat io n to h ome C. tr a cho ma ti s te sti n g w it h ref er ral to c lini c tes ti n g on C. tr a -c h o mat is p reva le nce and PID Mea n age , 1 8 yea rs ; 1 9 5 % w hit e w ome n Stu d e n ts at schoo ls ra n-dom ly assig n e d to screening w ere o ffered 1-t im e hom e samp lin g (vagi na l flus h fo r w o men) fo r C. tr a chom -atis NA A T (8 hig h schoo ls , 260 3 w o m e n ) a n d s tude nts a t c o n trol schoo ls we re giv en in -fo rmat io n a n d told c o uld visit clin ic fo r fr e e C. tr a c h omat is test in g (9 h ig h s c h ool s, 2 884 w o men) ; inc lude d o nl y sexua lly acti v e stu de n ts ; b o y s a ls o s c reene d Out c ome at 1 yea r: PI D inc id en c e , ascert a in e d b y s e lf-r epo rted treat -men t o r P ID hos pi ta li-zati on , con fi rm e d wi th p h a rmac y reg ist ry re -cord s (C . trac h o mat is pre v ale nc e re sul ts not p rese nt ed h e re) L e s s than h a lf re s p onde d to fi rst s u rve y (4 8 % in s c reen ing g roup a n d 3 8 % in c o n trol g roup) ; ∼ 75 % w e re sex ua lly acti v e and e n roll ed; 2 g roups s imil ar ; screening group: 867 (93%) of 9 2 8 e n roll ed w o men test ed fo r C. tr a cho ma ti s (5% pos it ive); c o n trol g roup : 6 3 (8 % ) o f 8 33 e n roll ed w omen test ed fo r C. tr a -c h o mat is (8 % p o s iti v e); ∼ 45% lost to follow-up at 1 year; out-come : PID in c ide n c e a t 1 yea r was 2. 1 % in scre en in g g ro u p a n d 4 .2 % in c o ntro l g rou p (P p .04 5) Randomized controlled d e s ig n ; d if fer ence s in rate s o f te s tin g in 2 g roups a llow s e va lu a -ti o n o f s c reeni ng L o w in itial p a rtic ipat io n, d if fere ntial e nrol lme nt b e tween g rou ps, a lmos t half los t to fo llo w -u p ; out c o m e asses smen t n o nbl ind ed; P ID o ut-come mea sur e d on ly by s e lf-r epor t a n d p h a r-macy re c o rd s ; find in g s fro m Dani s h hig h scho o l s tu dent s m ay n o t g en-e ra liz e to o ther p o pul at ion s S u b s ta n tial n umb er o f limi tat ion s, makin g it d if fi c u lt to dr a w firm conc lu sio n s , but sup p o rt s id ea th a t 1 -t ime p roac ti v e s c reen ing c a n reduc e P ID o v e r th e n ext year in high school students; m a g nit ude o f P ID p reve nt ed a p -pea rs hi gh e r th a n wh at mi g h t b e e x p ect ed b a s e d o n n a tural h is -to ry s tudie s; this a n d id e ntifi ed lim itat ion s rai se q uest io ns a s to w h ether fi ndi ngs a re g ene ral iz -able and pr o vid e rea listi c e x p e c -ta tions o f b ene fit in real -wor ld s e ttings
S159 Coho rt stu di e s w it h u nt est ed c o n trol g roup Clark et al [1 3 ] (20 02 ) N o nran d omi ze d c o h ort o f 2 8,0 74 femal e U S A rm y recr uit s in 1996– 1997, assessing e ffect o f C. tr a cho ma -tis screening on PID-re lat ed h osp ital iz a-ti ons th rough e n d o f 199 8 Over a ll, 80 % aged ! 25 years , 36% of wo me n wer e b lack Non ra ndo m ly assig ne d based on da y o f wee k to s c reeni ng g roup (o f-fe red 1 -tim e C. tr a -c h o mat is test ing b y u rine N A A T o n S u nday s at ba sic tr a ini n g in ta ke, 8 0 % v o lu ntee red (7 0 5 3 screened and followed; another 6151 screened wom e n ex clu de d be-cause no t e lig ib le fo r o n goi ng m ili tar y h ea lt h care) or control (not of-fe red C. tr a cho ma ti s tes ti n g if ar ri ved at ba -s ic trai ni ng o n o the r days of we e k [n p 21, 02 1 ]) Hos pi ta liza ti on via US Arm y adm in is tr a tiv e data , ICD-9 code s , all -cause and PI D, EP , a n d in fertili ty-rela ted, comp a ri n g scre en e d vs uns cre e n e d women; adjusted for age, race, education, and apt it u d e scor e S c reen ed g roup y o ung er , h igh er a p -titude score but less educated; C. tr a cho ma ti s pre v ale nc e 9.1 % in s c reen ed g roup; m e a n fol low -up pe ri o d of 1. 5 yea rs ; o u tc o m e o f P ID h o s p it a liza ti o n in s c reen-in g g roup w a s 4 .6 h osp ital iz a-ti ons p e r 1 0 0 0 P Y a n d in c o ntrol g roup w a s 5 .1 h osp ital iz a tions per 10 00 PY (R R, 0. 94 ; 9 5 % CI, 0.6 9– 1. 29 ); ho s pit a liz a ti o n s fo r PID, EP , o r inf ertil ity w a s 7 .2 h os-p ital iz a tions p e r 1 00 0 P Y in screening group and 6.8 h ospital-izations per 1 000 P Y in control gro u p (R R, 1. 10 ; 9 5 % CI , 0.8 5– 1. 43 ); ho s pit a liz a ti o n fo r any rea son wa s 199 hos pi ta liz ati o n s per 1000 PY in screening group a n d 2 2 4 h osp ital iz a tions p e r 1 00 0 PY in control group (RR, 0.94; 95% CI, 0.9 0– 0. 99 ); mo re pre g -nancy-related hospital stays in u n s c reen ed g roup Larg e sampl e siz e, pr o -spec ti v e col le c ti o n of data N o nran d omi ze d, p oten -ti a lly impo rtan t d if fer -e n c e s b e tween g roup s; no intention to screen analysis; n o o utpatient data or vali da ti on of PID dia gn os is A lt houg h n o nran do miz ed , this w a s a wel l-d o n e stu dy of no n-he a lth care– see k ing wo me n tha t ra is e s th e p o s s ibil ity th a t th e b e nefit s o f s c reeni ng m a y b e d if ficu lt to mea sur e in pra cti c e and de p e n -d e n t o n p opu lat io n; a ltho ugh lim -ited b y lack o f outpatient data, study suggests the ef fect of screening o n PID may b e w eaker th a n that o b s e rve d in the tria l b y Scholes e t a l [12] and Østergaard et al [11 ], de s p it e h ig h scre en in g cover a g e Low et al [1 7] (20 05 ) His to ri c a l coho rt of wom e n age d 1 5 – 24 years , Upp s ala C o u nty , S w eden (1 9 8 5 – 1 9 89) , follo wed u p throu gh 1 999 fo r o u tco me s rela te d to C. tr a cho ma ti s To tal o f 4 3,7 15 w omen , 52, 73 1 C. tr a cho ma ti s te s ts ; 2 0 ,8 5 3 w omen nev er te ste d fo r C. tr a -c h o mat is , 19, 89 7 al-ways C. tr a cho ma ti s neg a ti ve, 29 6 5 ev er C. tr a c h omat is p osi tive ; n o systematic screening, but 71 % e v e r te ste d : 48% te s te d on ce, 22% tes te d tw ice, 30 % tes te d ⭓ 3 times Regi s tr y data lin kin g C. tr a c h omat is te s t hi sto ry / re s u lts (1 la b orat or y; m o s tl y c u lture) w it h hos pi ta l ICD-9 cod es 198 5– 19 9 9 us in g u n iq u e ID ; a ls o , o u tpat ie nt d ia g -nos is code s fo r 199 3– 19 9 9 Cumu lati v e inc id en c e of PID, EP , infer tili ty; HRs com p ari n g ou t-come s b y C. tr a chom -atis te s ti n g sta tu s; ad-jus te d fo r age , edu c ati on , h o u sin g type , in c ome , n u m b e r o f b irth s, a n d c e n s u s year Among tested women, period prev-alen c e of C. tr a cho m ati s wa s 1 2 % ; c u mul at ive inci d ence b y 3 5 years of age fo r PID , E P, and in -fe rtili ty w a s 3 .9 %, 2 .3 % , 4 .1 %, res pe c ti vel y, amo ng all pa rt ici-pant s ; 5. 6% , 2 .7 % , and 6. 7% , res pe c ti vel y, amo ng C. tr a cho m a -tis – p osi tive w o men; 4 .0 % , 2 .0 % , and 4. 7% re spe c ti vel y, amo n g C. tr a c h omat is – n ega tive w o m en; and 2. 9% , 1 .9 % , and 3.1 % , re -spec ti v ely , amo n g wo m e n n o t tes te d ; com pa ri n g C . tr a c h oma-tis – p o s iti v e wit h C. tr a cho m a -tis – n e g ati v e wo me n , PID (aHR , 1.3 ; 95% CI, 1. 0– 1.6 ), EP (aHR , 1 .3 ; 9 5 % C I, 0 .9 – 1 .7 ), infer tili ty (aHR , 1 .3 ; 9 5 % CI, 1. 1– 1. 6); nev er te ste d w o m e n les s likel y tha n C. tr a cho m ati s – n eg at ive w o men to h a v e P ID o r infer tilit y dia gn os is P o p u lat io n-b as e d c o hor t d e s ig n w ith g ood rec-ord link age , lon g fol -low -u p , la rg e samp le size; inc lu de s w o m e n w h o w ere n eve r tes te d fo r C. tr a c h omat is M o s t w omen h a d o n ly 1 – 2 te s ts over 10– 14 years; thus, undetected in fect io ns a t o ther ti m e s wer e likel y, es pec ia lly among those with ⭓ 1 d e tect ed C. tr a cho m ati s in fect io n; le s s s e nsi tive tes ts us ed ; coul d n o t c o n trol fo r s e x u a l b e -havior , contraceptive use, etc; infertility o ut-come onl y availa bl e fo r wom e n see k ing tr e at-men t; un abl e to d if fe r-e n tiat e femal e a n d m a le factor infertility Onl y re g istr y stu d y wi th an un-te s te d c o m par is o n g roup; s h o w s th a t, o u tsi de a trial , w om en w h o have no t b e e n te s te d are likel y to b e d if feren t, a n d p ro bab ly a t low e r ris k o f C. tr a cho m ati s than are those who have been tested; u n det ect ed, u n treat ed infec tions a t o ther ti m e s w ou ld tend to re -duc e obs er v e d im p act of te s ti n g o n la te r long -t erm s e q uela e ; b en-efit s o f scre en in g d if fi c u lt to as-sess us in g th e s e ty p e s of ob s e r-vational data; study suggests th a t w omen w h o h av e rec e ive d a dia gn os is of C . trac h o mat is in -fec ti o n are mo re likel y to see k care for PID, E P, and infertility
S160 Coho rt stu di e s w ithout u n -te s te d c o n trol g roup A n d erse n e t a l [14 ] (20 05 ) His to ri c a l coho rt of wom e n age d ! 43 years te s ti n g fo r C. tr a c h omat is in 1 coun ty in De nm a rk (19 84 – 199 3 ); N o sys-te mat ic s c reenin g, te s t u ptak e n ot re -p o rted. te s ts d one mainly for symptoms, before transcer vical procedures, for part-ner no ti fi c ati on , and opportunistic screen-in g in s o m e s e ttin gs (eg , STD cli n ic s ) 1 3 ,6 9 3 w omen, 1 8 ,4 5 2 tes ts do n e be fo re fir s t pre g n anc y Regi s tr y data us in g u n iq u e ID to link C . trac h o mat is te s t resu lt s (1 lab or a-tor y, all pr e-N A A T ) w it h pre g n anc y o u tcom e b y hos pi ta l d iagn o sis cod e s E P inci d enc e rat e, H R c o m par in g C. tr a -c h o mat is – p osi tive wit h C. tr a cho m a -tis – n ega tive w o m en; exclu d e d w o m e n wi th n o p reg nan cy th rough 1 Janu a ry 200 3 (fin al, 607 2 w o m e n); bi rt h rate , H R fo r C. tr a -c h o mat is – p osi tive v s C. tr a cho ma -tis – n ega tive w o m en; in c lu d e d w omen w h o did not be com e pre g -nant through age 43 years or 1 Janua ry 2 0 03; a d just ed fo r m a ternal a g e C. tr a cho ma ti s period prevalence 1 4 % ; n ear ly a ll p a tient s rece iv e d a p p ropr iat e trea tment ; E P inci -d e nce ra te lo w e r in w omen w ith ⭓ 1 p osi tive C. tr a cho ma ti s te st res ul t: 0. 35 cases /10 0 P Y amo n g C. tr a cho ma ti s – p o s itiv e w ome n, 0.8 6 cases /10 0 P Y amo ng C. tr a -c h o mat is – n e g ati v e wo me n (aHR , 0.5 5; 95 % CI, 0.3– 0. 96 ); ti me to birth similar in b oth groups; birth rate amo n g C. tr a cho m a -tis – p osi tive w o men w a s 5 .8 4 b irths/ 1 0 0 P Y a n d a m ong C. tr a -c h o mat is – n e g ati v e wo me n was 5.9 6 b ir th s /10 0 P Y (a HR, 0.9 2; 95% CI, 0.8 4– 1. 00 ) P o p u lat io n-b as e d c o hor t d e s ig n w ith g ood rec-ord link age , 1 10 yea rs o f follo w-u p, la rg e sampl e siz e Rate s amo n g wo m e n nev er te ste d fo r C. tr a -c h o mat is no t evalu a te d ; no syste m ati c scre e n -ing , le s s sen s iti v e te sts used; u ndetected infec-tio n s at oth e r ti me s wer e likel y in b o th groups, especially for tho s e wi th ⭓ 1 d e tect ed C. tr a cho ma ti s in fect io n; could n ot control for so-c io dem ogr ap hic fa c tors, sexual behavior , contra-c e p tive u s e , d esi re to conc ei ve, et c Stu d y do e s no t asses s ben e fi ts of screening; undetected, u ntreated in fect io ns a t o ther ti m e s (i n e i-th e r g rou p) m ig h t a ffec t inter pre-tation; birth rate may be slightly low e r in C. tr a cho m ati s – p os itive g roup, b u t d o n o t k n o w w hethe r factors, such a s d esire to con-ceive, dif fered in the 2 g roups; canno t nec es sari ly ex tr a pol a te birth rates to infertility Bakken et al [15 ] (20 07 ) His to ri c a l coho rt of w o men b orn d urin g 197 0– 19 8 4 te s te d fo r C. tr a cho ma ti s in 1 coun ty in No rw a y (19 90 – 200 3 ); n o sys-te mat ic s c reenin g; re -p o rted test ing ra te s h igh; ∼ 80 % o f w o men test ed, 1 50% had 1 1t e s t 2 0 ,7 6 2 w omen, 7 2 ,4 0 5 tes ts do n e be fo re fir s t pre g n anc y Regi s tr y data lin kin g C. tr a c h omat is test re s u lts (1 la b orat or y, 5 1 % N A A T ) w ith in pat ie nt and ou tp a ti e n t E P code s and bi rt h s EP in c ide n c e ra te (HR c o m par in g C. tr a -c h o mat is – p osi tive wit h C. tr a cho m a -tis – n ega tive w o m en; evalu a te d wh o le gro u p and lim it e d to wom e n wi th EP or b irth in s tudy p erio d [n p 10, 79 4 ]); bir th rate (H R fo r C. tr a -c h o mat is – p osi tive v s C. tr a cho ma -tis – n ega tive w o m en) ; cens or e d at fir s t bir th or 31 Dec em be r 2 0 04; a d just ed fo r pari ty and age at firs t C. tr a cho ma ti s te st C. tr a cho ma ti s period prevalence 18% ; h ig h e r EP ra te s amo ng C. tr a c h omat is – p o s iti v e wo me n v s C. tr a cho ma ti s – n e gat iv e w ome n in ov era ll pop u la ti o n and lim it e d to those w ith p regna n c ies; d o s e res po n s e w it h nu m b e r of pos i-tiv e C. tr a cho m ati s test re s u lts; among those with pregnancies: EP in c ide n c e amo ng wo m e n w it h n o p osi tive te s t resu lt s w as 0.29 cases/100 P Y (reference), among w omen with 1 p ositive tes t re s u lt wa s 0 .5 8 cases/1 00 PY (a HR, 1.8 ; 95% CI, 1. 1– 3.0 ), and amo ng wo me n w it h ⭓ 2 pos -iti v e te s t re s u lt s w a s 1. 39 cases/ 100 PY (a HR, 3.4 ; 9 5 % CI, 1.5– 8. 0); bi rt h ra te sli g h tl y hi gh e r amon g C . trac h o mat is – p osi tive w o men (1 0 .2 b irth s/ 10 0 P Y ) th a n amon g C . trac h o mat is – n egat iv e w o men (9 .6 b irths /100 P Y ; a H R , 1.0 7; 95 % CI, 1.0 1– 1. 12 ) P o p u lat io n-b as e d c o hor t d e s ig n w ith g ood rec-ord link age , lon g fol -low -u p , la rg e samp le size; mor e comp le te te s ting, m o re N A A T te s ting than o ther Scand in a vian re g is tr y s tu dies ; in pat ie nt a n d outpatient data a p lus Rate s amo n g wo m e n nev er te ste d fo r C. tr a -c h o mat is no t evalu a te d ; no syste m ati c scre e n -ing , cha nge in te s t te c h -n o log y o ve r time; u n d e -te c te d in fect io ns a t oth e r ti me s w e re likel y in bo th gr o u p s , es pe -ciall y fo r th o s e w it h ⭓ 1 d e tect ed C. tr a cho m ati s in fect io n; c o u ld n o t c o n -tr o l for s o c iode mo-g ra p hic fa c tors, s e x u a l behavior , contraceptive use, de s ire to conc ei v e , etc Stu d y do e s no t asses s ben e fi ts of screening; compared w ith D anish stud y, hi g h e r ri sk of EP amon g tho s e te sti n g po s iti v e and tr eat ed fo r C . trac h o mat is infec -ti on; d o s e resp o nse s u g gest s re-c u rrent in fect io n in c reas e s c u mu-la tive risk o f E P ; u n detec ted, u n treat ed in fect ion s a t o the r ti mes (i n e ithe r g rou p) m ig h t a f-fe c t inte rpretat ion; a s a b o v e , d if-ficul t to e x tr a p o la te bir th ra te s to in ferti lit y
S161 Hil lis e t a l [16 ] (19 97 ) His to ri c a l coho rt stu dy o f n umb er o f C. tr a -c h o mat is infec tions and en sui n g ho s pit a li-zati on s fo r EP and PID (PI D res ul ts not p rese nt ed h e re) ; W is -cons in (1 98 5 – 199 2 ); mand a te d ann ua l ri sk-based scre en in g in fa m ily p lan ni ng c lini cs , uni v ers al scre e n in g in 2 S TD c lin ics 1 1 ,0 0 0 w omen a g e d 10– 44 yea rs wi th ⭓ 1 C. tr a c h omat is infe ct ion; w o men w ith ⭓ 3 infec -ti ons: 9 7 % a g e d ! 25 years ; ∼ 5 0 % b lack, ∼ 50 % N. go no rr h o e a e c o infec tion a t fi rst in fect io n C. tr a cho ma ti s in fect io ns re ported b y p u blic p ro-vide rs in W is cons in C. tr a c h omat is case re g is -tr y 1 9 8 5 – 1 9 9 2 (a ll no n-NAA T), lin k e d b y n a me, date of birth, and c o u nty w it h s ta tewid e hos pi ta l d ischa rg e IC D-9 data ba se 198 9 – 199 4 (n o in forma tion b efor e 1 9 89) Comp a re d EP hos pi tali -zati on s amo ng th os e wit h ⭓ 3 C. tr a cho ma -tis in fect ion s (n p 6 4 4 ) w ith those w ith 2 C . trac h o mat is in -fe c tions (n p 20 4 4); rand o m samp le of w o men w ith 1 C. tr a -chomatis infection (n p 831 2 ); adj us te d fo r N. g o norrhoea e infec tion, coun ty , sou rc e o f care C. tr a cho ma ti s period prevalence 11% (fa mi ly pl a nni n g), 13 % (S T D clin ics). 98 % o f re p e a t C. tr a -c h o mat is d iag n ose s ⭓ 3 0 da ys apart ; per c e n ta g e w it h E P hos pi -ta liza tion (1 989 – 1 994 ) for those wit h 1 , 2 , and ⭓ 3 inf ect ion s w as 0 .1 3 % , 0 .4 9%, 1 .4 % , resp ec-tiv el y, amo ng all wo m e n , and 1.3 %, 2.5 %, and 4. 8% , re s p e c -tiv el y, amo ng wo me n w it h pre g -nancy; compared w ith those with 1 C . trac h o mat is infec tion, e le -vated odds of EP among women with 2 infections (OR, 2.1; 9 5% CI, 1. 3– 3. 4), ⭓ 3 infec tions (O R , 4.5 ; 95% CI, 1. 8– 5.3 ), N. go n o r-rh oeae /C . tr a cho ma ti s c o infect io n (OR, 5.3 ; 9 5 % CI, 1. 9– 15 .0 ) P o p u lat io n-b as e d c o hor t des ig n, la rg e samp le size No com pa ri s o n g ro u p (C. tr a c h omat is n ega tive o r nev er te ste d ); n o p e r- son-ti m e ana ly sis; hos pi -tal di scha rg e data lik ely to be fa irl y accur a te fo r EP , but no da ta be fo re 1 9 8 9 a n d n o o utpa tient data in late r yea rs ; lim -it e d in forma tion o n p o -te ntial c o nfou nde rs Stu d y do e s no t asses s ben e fi ts of screening; the risk of EP among th ose w it hout a p as t p os itive C. tr a c h omat is te s t re sul t o r amo n g th ose n e v e r te s te d u nk n own; d o s e res p onse s u gge st s recu r-ren t in fe c ti o n in cre a ses cumu la -tiv e ri s k o f E P Ecol og ical stu d ie s Egg er et al [2 1] (19 98 ) Ecol og ical stu d y of tr ends in ra te s o f C. tr a c h omat is infe ct ion and EP , U p p s ala C o u nty , S w eden (1 9 8 5 – 1 9 95) ; 1 03,8 75 c e rvi ca l s a m ples a n d 5 1 ,6 3 0 p regna n c ie s analy zed ; n o syste m -atic screening; oppor -tu nist ic test in g re-p o rted to b e inten si ve W ome n age d 20– 39 yea rs (same po p u la ti o n a s [17 ]) Coll e cte d C. tr a cho ma ti s test reports from single la b orat or y, h o s p it a l d is -charg e ICD-9 code s Rate s o f C. tr a cho m ati s dia gn os is per 10 0 tes ts , rate s o f E P p e r 1 0 0 0 p regna n c ie s , b y age gr o u p ; Poi s son regression used to exami n e assoc iati o n b e tween risk o f E P and cur re n t ra te s of C. tr a cho ma ti s and rate s o f C. tr a cho m a -tis fr o m up to 5 yea rs earl ie r Over a ll C. tr a cho m ati s test p o s itiv -ity 5.4%; 1 .8% o f pregnancies wer e EP; bo th C. tr a cho m ati s and EP ra te s d e c re ased mark -edly over time; among women aged 20– 24 years , th e re wa s a s tr ong c o rrelat io n b e tween E P rate and C. tr a cho ma ti s ra te in same yea r (r p 0. 93 ; P ! .0 01) ; among o lder w omen, the correla-ti ons w e re le s s p rono unce d , b u t mor e str o n g ly assoc ia te d w it h C. tr a c h omat is ra te s 1 o r 2 yea rs earl ie r Pop ul a ti o n -b a sed ascer -ta inmen t o f E P rat es d u ring a p er iod o f in -te nsi ve s c reeni ng; larg e sampl e siz e, 10 years of da ta Ecol og ical de sig n ; th us , n o conc lu sio n s can be dra wn abo ut caus alit y ; in div idua l b en efits o f screening cannot b e as-sessed ; m a y be un m e a -sured confounders, such as intrauterine de-vice us e Among young women, strong cor -re lat io n b et ween C . trac h o mat is and EP and con com it ant de -crease in rates supports C. tr a -c h o mat is as caus e o f E P ; de -crea ses in C . trac h o mat is and EP occur red during same period a s C. tr a cho ma ti s c o n trol e fforts, but also d uring same period a s o ther p o tent ial ly c o nfou ndin g fac -tor s (e g , in c re ases in safe r sex w it h H IV e pid emi c, b etter d iag -nosis and treatment of symptom-atic C . tr a c h omat is infe ct ion, d e -crea ses in N . g ono rrhoea e in fect io n Hil lis e t a l [18 ] (19 95 ) Ecol og ical stu d y of tr ends in C. tr a cho m a -tis case and la bo ra -to ry repor ts (19 85 – 199 1 ) and ho s -pit a liz ed PID and EP (19 82 – 199 1 ), W is con -sin; sta te C. tr a cho m -atis p reve n tion p ro-g ra m ini tia te d in 1 985 , inc lu di n g C. tr a cho m a -tis s c reeni ng in fa m ily pla nn in g and STD clin ic s … C o m bin ed d a ta from W is-cons in ST D Sur v e ill a nce Syste m and pub lic h e a lt h lab or at or y d at a-bases , alo n g wi th W is -cons in Ho s pit a l Dis -charg e Data ba se (be fo re 1 9 89, p a rtici pa tion v o lunt ar y) R e p orted C. tr a cho m a tis rate s p e r 100 ,0 0 0 p o pul at ion , p ropor tion of C . trac h o mat is tes ts po s iti v e fo r C. tr a c h omat is , tr ends in PID and EP ho spi ta li-zati on ra te s Among w omen aged 1 5– 24 years, re ported C. tr a cho m ati s in fect ion rate s in c rea sed sharp ly , th e n le v -e le d o ff , b ut p roporti on o f test s p o s itiv e for C. tr a cho ma ti s ste ad-ily de cre a sed ; amon g all wo m e n (age d 15– 44 yea rs ), ra te s o f PID hos pi ta liz ati on de cre a sed by 31 % d u ring 1 9 8 2 – 1 98 5 a n d 3 3 % d u r-in g 1 9 8 6 – 1 991; E P ra te s in-crea sed by 14 % d u ri n g 198 2– 19 8 5 and dec re a sed by 20% du ri n g 19 8 6 – 1 9 9 0 ; n o sta -ti s tics re p orted fo r tren ds Bro ad asses smen t o f statewide trends in re lat io n to im ple men -tati o n of an in te n s ive C. tr a cho ma ti s pre -v e n tion p rog ram Ecol og ical de sig n ; th us , n o conc lu sio n s can be dra wn abo ut caus alit y ; in div idua l b en efits o f screening cannot b e as-sessed; n o o utpatient data , espe c iall y imp o r-tant fo r PID; no sta ti s-ti c s repor ted fo r trends D e c reas e s in p ropor tion o f C. tr a-c h o mat is test s p os itive a n d in E P a n d P ID ra te s o cc u rred d u ring same pe ri o d as C . trac h o mat is c o n trol e fforts, a lt h ough P ID rate s app e are d to b e d e cre a sin g b e fore C . trac h o mat is e fforts started; other potentially con-fo undi ng fa c tors a ls o o cc u rred in same pe ri o d (e g, inc re a ses in safer sex with HIV e pidemic, b e tter d ia g n o s is , a n d tr eat ment o f s y m ptom at ic C. tr a cho ma ti s infection); d ecreases in N. gon o r-rh oeae infec tions ; c h a n g e to o ut-p a tient m a n a g e men t o f P ID
Kamw en d o et al [1 9] (1 99 6 ) Ecol og ical stu d y of tr ends in P ID h osp ital -izations, N. go no r-rh oeae – and C. tr a -c h o mat is – assoc ia te d PID, and N. gon o r-rh oeae and C. tr a -c h o mat is case re p o rt rate s , O¨ rebr o, S we-den (19 70 – 1 9 9 4 ); p u bli c h ea lt h o ffici al s b e gan e nco ur ag ing d i-a g n osi s a n d trea tment of C . trac h o mat is a ro und 1 984, a n d C. tr a c h omat is be c ame a not ifi abl e d is ease in 198 8 … Evalua te d hos pi ta l re c ord s of pa ti e n ts wi th acut e PID in O¨ re br o fo r N. g o norrhoea e /C. tra-c h o mat is la bo ra to ry re -cord s : N . g ono rrhoea e te s ting 1 970 – 1 994 (n p 2 4 99) , C . tr a c h omat is te s ting 1 981 – 1 994 (n p 1 0 30) P ropo rtion o f P ID w ith C. tr a cho ma ti s /N. g o norrhoea e ; PID in ci-den c e , usin g pop u la -tio n sta tis ti c s of ho s -pital catchment area; N . g o norrhoea e and C. tr a cho ma ti s inc i-den c e ba sed on la bo-rator y reporting P ropo rtion o f P ID w ith C. tr a chom -atis de c re ased fr o m 28 % in 1 9 8 5t o 8 %i n1 9 9 4 ; N . g onor -rh oeae in 42% of PID in 197 0 , w it h d ec reas e to 0 % s ta rting in 1 9 80; P ID in c idenc e d ec reas e d cons is te nt ly fr o m a p e a k in 1 9 7 5 – 1 979 throug h 1 99 0– 1 9 9 4 fo r th e y o u nge st a g e g roups (15– 19 and 20 – 2 4 yea rs ); PI D re l-ativ el y sta bl e in o ld e s t age g roups ( 1 30 yea rs ); C. tr a cho ma -tis case re p o rt ra te s dec re a sed 40%– 60% d uring 1 988– 1994 Comprehensive assess-men t o f acut e PID hos pi ta liz ati on s and la b orat or y test in g in a defi n e d catc hm en t area ov er 25 yea rs Ecol og ical de sig n ; th us , n o conc lu sio n s can be dra wn abo ut caus alit y ; in div idua l b en efits o f screening cannot b e as-sessed; n o o utpatient data , espe c iall y imp o r-tant fo r PID; no sta ti s-ti c s repor ted fo r trends Dec re a ses in C. tr a cho ma ti s , PID , and C. tr a cho m ati s – re late d PID occur red during same period a s C. tr a cho ma ti s c o n trol e fforts, a l-th ough P ID ra te s a p p e a re d to b e decreasing before C. tr a cho ma ti s e fforts s ta rted ; o ther p o tent ial ly c o n foun din g fac tors a ls o o c-curred in same period (eg, in-crea ses in safe r sex wi th HI V e p -id emi c, b etter d iag n osi s , a n d treatment of symptomatic C. tr a -c h o mat is infec tion) ; d ec reas e s in N . g o norrhoea e infec tions ; change to o utpatient manage-men t o f P ID Chen et al [20 ] (20 06 ) Ecol og ical stu d y of tr ends in c lin ica l e n-c o u nter s for P ID in gen e ra l pr a cti ce and C. tr a cho ma ti s case report rates, Australia (1 9 9 8 – 2 0 03) ; e xt ent of C . trac h o mat is con -tr o l e ff orts in a s imil ar p e riod n o t report ed … Comp a re d da ta fr o m na -ti onal ly repr ese nt at ive sur v e y o f Austr a lia n gen e ra l pr a cti ces on pa -ti ent e n c o u n ters (5 0 2 ,1 0 0 e n c o u n ters 1 9 9 8 – 2 003) w it h n a -ti onal C . trac h o mat is case report d ata Rate of PID di a gno s is per 10 ,0 00 fe m ale pa-ti ent e n c o u n ters, re -p o rted C . trac h o mat is rate s p e r 100 ,0 0 0 p o pul at ion Among w omen aged 1 5– 34 years, PID en cou n te r ra te s dec re a sed fro m h ig h o f 3 9 .9 cases /10 ,0 00 e n c o u n ters in 1 999 – 2 00 0 to a low of 19. 4 cases /10 ,0 0 0 en -c o u nter s in 2 0 01– 2 0 02; m e a n dec re a s e o f 3 .6 cases /10 ,0 00 en -counters p er year (P p .02 ); no change in rate o f PID encounters in wo m e n aged ⭓ 3 5 yea rs ; C. tr a c h omat is case re p o rt s in -crea sed fr o m a ra te o f ∼ 50 cases/1 00 ,0 0 0 po p u la ti o n in 1 9 9 4 to ∼ 180 cases/100,000 popula-tio n in 2 0 0 3 ; pr e c ise fig ur e s not re ported U s e o f o u tpat ie nt P ID data from nationally re prese n tat ive assessme n t Ecol og ical de sig n ; th us , n o conc lu sio n s can be dra wn abo ut caus alit y ; in div idua l b en efits o f screening cannot b e as-sessed ; o n ly C. tra -c h o mat is case re p o rt s used, n o information on tes t p o s iti v ity ; e x ten t o f C. tr a cho ma ti s c o n trol e fforts n o t repor ted M o re rece nt s tudy in c lu d ing o utpa -tie n t data sho w s th at PI D ra tes appe a r to be dec re a sin g in you n g w o men; u n c le a r w hat h a s o c-c u rred in s im ilar ti m e w it h re-spec t to C. tr a cho ma ti s c o n trol activ it ie s , C . trac h o mat is p reva -len c e , and in cid e n ce; C. tr a chom -atis case re p o rt s ma y re fl e c t in-crea se in te s ti n g NOTE. Data from the Prevention of Pelvic Infection (POPI) trial, an additional randomized controlled trial (RCT) of C. trachomatis screening to prevent PID, were published too late to be included in this review but are now available [32]. aHR, adjusted hazard ratio; CI, confidence inter val; ELISA, enzyme-linked immunosorbent assay; HMO, health maintenance organizat ion; HR, hazard ratio; ICD-9, International Classification of Disea ses, Nint h Rev ision ; NA A T, n u c lei c acid ampl ific ati on te s t; O R , odd s ra tio; PM , p e rs o n -m on th s ; PY , p e rs on-y e a rs ; RR, re lative ri sk; STD , sex ua lly tr ans mi tte d dis ease.
Screening and Treatment to Prevent Sequelae • JID 2010:201 (Suppl 2) • S163
were ultimately enrolled. Women aged 18–34 years who were registered with the HMO were randomized to screening and control groups and then sent a survey to determine study el-igibility based on a risk score (Table 1). The authors reported more aggressive contacting of survey nonresponders from the intervention group to expedite testing appointments after eli-gibility determination. This may have resulted in the observed 1:1.6 ratio instead of the expected 1:2 ratio between the in-tervention and control groups, introducing the possibility of selection bias. It is not clear how such a selection bias might affect estimation of screening benefits, but randomization may have been compromised.
Second, by applying the rate of PID in the control group to person-months in the intervention group, we would expect 21 cases of PID, but only 9 cases were observed. Thus, 12 cases of PID were apparently prevented by treating 44 women, sug-gesting that at least 27% of the infections identified by screening would have progressed to diagnosed PID without any inter-vention. This figure appears substantially higher than findings from other studies of asymptomatic populations undergoing screening who were followed up untreated for 12 months [3, 8]. In addition, C. trachomatis has been implicated in∼30% of acute PID cases [23]; thus, even if all chlamydia-associated cases of PID were prevented by screening, a 50% decrease is larger than what might be expected on the basis of previous studies.
Østergaard et al [11] conducted a cluster randomized trial of 17 high schools in a Danish county, in which students in intervention schools were offered 1-time screening of home-collected specimens for C. trachomatis and students in control schools were given referral information about clinic-based test-ing. Less than half of the students returned eligibility surveys. Because 93% of participating students in the intervention group had C. trachomatis testing performed, compared with only 8% of control students, the effect of screening could be evaluated. Female students were followed up in 1 year, and incident PID was assessed by self-report, with confirmation by pharmacy records. Outcome assessment was unblinded, and almost 50% of students were lost to follow-up; thus, this study had a sub-stantial number of limitations. Nonetheless, it had findings similar to those from the trial by Scholes et al [12]. The 1-time screening approach in the Danish study was associated with a halving of PID occurrence over 1 year (4.2% vs 2.1%; P p ). Testing of 867 female students and detection of 43 in-.045
fections prevented 9 reported cases of PID among the 443 interviewed students at follow-up. Unlike Scholes et al [12], Østergaard et al [11] also screened boys attending the same schools, and this may have had an impact on community trans-mission, preventing some new infections and reinfections that could have also led to PID.
Cohort studies. An impact of the magnitude found in these
2 randomized controlled trials should be easy to detect. How-ever, a large cohort study of chlamydia screening in female US Army recruits by Clark et al [13] did not reproduce these findings. In this study, a total of 7053 women participating in a screening program on Sundays at a basic training intake center were tested and treated for C. trachomatis infection, and 21,021 women who arrived on different days were not. These women were followed up for hospitalizations over a mean of 1.5 years with use of US Army coded administrative data. There were no significant differences between the screened and unscreened groups in hospitalization rates for chlamydia-related outcomes (PID, infertility, or ectopic pregnancy). This study was not randomized and was limited by lack of outpatient data, which is a large concern because, at the time of the study, most PID cases were treated in the outpatient setting [24]. However, it does raise the issue that the positive effects of screening may be difficult to ascertain and dependent on the population, and suggests the effect may be weaker than that observed by Scholes et al [12], at least in some populations.
The 2 randomized controlled trials and the study by Clark et al [13] used symptomatic, clinically diagnosed PID as the study outcome. No randomized trials have directly evaluated the effect of chlamydia screening on ectopic pregnancy and infertility, conditions that may not be observed for several years, in part because young women’s use of contraception to prevent unplanned pregnancy delays their diagnosis. Three historical cohort studies in Scandinavian countries and a study in Wis-consin used clinical and population registers to link C.
tra-chomatis testing records with data on long-term complications
[14–17]. These studies, the findings of which are summarized in Table 1, yield useful information on the risk of long-term complications in women with at least 1 diagnosed and treated
C. trachomatis infection but provide little insight on the role
of screening in preventing these complications.
Only one of the historical cohort studies, the Swedish study by Low et al [17], had information about a comparison group of women who had never been tested for chlamydia. This group had lower rates of PID and infertility diagnoses than did women who were tested, including women who only had negative chla-mydia test results. This highlights the fact that, outside a ran-domized trial, women who have not been tested are likely to be different and probably at lower risk of chlamydial infection, compared with those who have been tested. Another meth-odological issue affecting interpretation of these studies relates to undetected infections. The intensity of C. trachomatis testing and control has been greater in the 3 Scandinavian countries in which the cohort studies were done than in most other European countries [2]. Nonetheless, data reported in the Nor-wegian and Swedish studies show that most women had only 1–2 tests performed over a 10–14-year period [15, 17]; there-fore, there is a high likelihood of undetected infections at other
S164 • JID 2010:201 (Suppl 2) • Gottlieb et al
times. Because C. trachomatis is associated with high rates of repeat infection [25, 26], women with 1 detected chlamydial infection are particularly likely to have had ⭓1 other unde-tected, untreated infection. This would tend to reduce the ob-served impact of testing on subsequent long-term sequelae. All in all, it is difficult to make an unbiased assessment of the benefits of screening using these types of observational data.
Ecological studies. Ecological studies have been cited as showing the effectiveness of screening programs, based on co-incident decreases in chlamydia and chlamydia-related out-comes observed in regions that were the earliest to initiate chlamydia control efforts (Table 1) [18, 19, 21]. During the 1980s and early 1990s, when chlamydia control activities were being implemented, rates of identified chlamydial infection de-creased in conjunction with rates of PID [18, 19] and ectopic pregnancy [18, 21]. However, in some areas, PID rates appeared to be decreasing before the chlamydia control programs started [18, 19], and many factors, such as decreases in the number of gonorrhea infections, increases in safer sex with the human immunodeficiency virus (HIV) epidemic, and better diagnosis and treatment of symptomatic C. trachomatis infection, oc-curred at a similar time. Since the late 1990s in Australia, out-patient clinical encounter rates for PID have continued to de-crease among young women, whereas rates of reported chlamydia cases have increased substantially [20]. Similar find-ings have been observed in British Columbia [27]. Chlamydia control programs may well be contributing to ongoing de-creases in chlamydia-associated sequelae, but ecological studies cannot be used to determine causality.
DISCUSSION
Summary of existing data. Only a few studies have evaluated the direct benefits of screening for C. trachomatis genital in-fection. Two randomized controlled trials of 1-time screening for chlamydial infection—in a Seattle-area HMO and a Danish school district—revealed an ∼50% reduction in the incidence of PID among screened women in the following year. However, both of these trials had methodological issues that may have affected the magnitude of observed screening benefits in the populations evaluated and that might limit the generalizability of these findings to real-world settings. A large, nonrandomized cohort of chlamydia screening among US Army recruits did not find a substantial reduction in the number of hospitali-zations for PID. Data from historical cohorts have methodo-logical limitations that restrict the capacity to make valid con-clusions about the benefits of screening. Ecological studies have generally supported the effectiveness of chlamydia control pro-grams in decreasing population-wide sequelae. However, be-cause of coincident temporal changes in behavior, clinical prac-tices, and rates of other infections, no firm conclusions can be drawn about the role of chlamydia screening in reducing
com-plications, and it is impossible to assess benefit on an individual level.
Gaps in knowledge. At present, it is unknown whether findings similar to those observed in the randomized controlled trials reviewed here would be found in other populations pos-sibly at lower risk, and it is also unclear whether the screening efforts implemented in these trials apply to screening programs currently in place. Perhaps most important, major gaps in knowledge include whether screening and treatment of asymp-tomatic, prevalent infection can prevent long-term sequelae, such as infertility and ectopic pregnancy—the outcomes we most want to prevent. Most of the studies we reviewed evaluated only 1-time screening, and there is little understanding of the overall benefits to an individual of11 round of screening, the
optimal frequency of screening, or the benefits of screening for recurrent infections [10]. In addition, the relative benefits of different types of screening approaches to prevent sequelae have not been evaluated. Finally, development of optimal screening strategies is limited by gaps in knowledge related to the natural history of C. trachomatis infection. For example, understanding the benefits of screening for infertility prevention depends in part on whether the pathologic processes leading to symptom-atic PID (the main outcome in available trials) are the same as those leading to long-term outcomes. The effectiveness of a screening strategy also depends on the risk and timing of tubal damage relative to acquisition of infection and the mean du-ration of infection in the targeted population. Inaccurate mea-surement of the outcomes of chlamydial infection hampers our ability to evaluate screening strategies in both research and nonresearch settings.
Implications. Direct evidence about the effectiveness of finding and treating prevalent, asymptomatic chlamydial in-fection in preventing adverse sequelae clearly has major im-plications for chlamydia control efforts. Sequelae of chlamydial infection in a population can be prevented either by curing existing infections before they progress to PID and/or tubal damage or by preventing new C. trachomatis infections [28]. How a chlamydia control program should be structured, thefore, depends on the relative effectiveness of screening in re-ducing sequelae in infected women, compared with interrupt-ing transmission and, thereby, reducinterrupt-ing incidence in the population. Chlamydia screening has often been perceived as providing its main benefit through identification of women with chlamydial infection early enough to treat them and pre-vent progression to sequelae in those individuals [22]. This is the type of benefit that was evaluated in the trial by Scholes et al [12]. Until recently, cost-effectiveness analyses evaluating chlamydia screening as a way to prevent PID were structured around halting disease progression in already infected women rather than using screening as a tool to decrease transmission and prevent new infections in a population [29, 30]. A program
Screening and Treatment to Prevent Sequelae • JID 2010:201 (Suppl 2) • S165
focused primarily on reducing the number of sequelae in in-fected women makes the most sense if women with asymp-tomatic, prevalent infections still receive a substantial amount of benefit from treatment (ie, if tubal inflammation and damage would continue to be elicited beyond the time that a typical infection is detected by screening). However, if screening has little impact on preventing sequelae in infected women, because tubal damage, if it is to occur, happens relatively soon after the infection is acquired, control programs must focus primarily on reducing incidence of new chlamydial infection in the pop-ulation. A program to reduce incidence of new infection in a population through interruption of chlamydia transmission might put greater emphasis, for example, on treatment of sex partners. Thus, more-precise estimates of the benefits of chla-mydia screening to an individual infected woman may lead to consideration of new ways to optimally restructure chlamydia control programs to reduce adverse outcomes of chlamydial infection in a population.
Next steps and research needs. Additional studies of the effectiveness of detecting and treating asymptomatic, prevalent chlamydial infection in preventing sequelae would be useful. For countries that do not currently have an active chlamydia screening program in place as the standard of care, randomized controlled trials to better delineate the individual benefits of screening [12] could be done and would be valuable. Before the National Chlamydia Screening Programme was fully estab-lished in England, a randomized controlled trial of 1-time chla-mydia screening among college-aged women was conducted and recently completed [31]. Published data from the trial were released too late for inclusion in this review but are available now [32]. Ideally, screening studies would collect data in a way that does not focus solely on prevention of PID in screened individuals, but also includes an assessment of community-wide benefits. Screening a group of women could reduce trans-mission in a population and, thus, prevent new infections, with their own attendant risk of sequelae, in women in the whole population.
Countries considering implementation of chlamydia screen-ing programs have the opportunity to use randomization ap-proaches, such as the stepped wedge design, to systematically evaluate the effects of screening, as is currently being done in the Netherlands [33]. In areas where screening for C.
trachom-atis infection is already recommended, assessments of the
ben-efits of screening may be more challenging. However, additional creative approaches that make comparisons according to changes in screening intensity and coverage would add insight. Community randomization to evaluate the benefits of enhanced chlamydia screening efforts is currently being undertaken in Australia and provides an opportunity to collect valuable in-formation (J. Hocking, personal communication). However, potential methodological problems need to be anticipated,
in-cluding underlying baseline differences in community chla-mydia prevalence, achieving high screening coverage, testing and treatment of sex partners, and accurate, unbiased ascer-tainment of outcomes.
A better understanding of the natural history of C.
trachom-atis infection is also essential to improving and informing
chla-mydia control efforts [3]. Ideally, natural history studies would help better clarify the incidence and timing of PID and tubal damage leading to long-term sequelae after untreated chla-mydial infection. Such assessments would need to be done in different populations, including asymptomatic women with prevalent infection who have no indication for testing other than screening. Because it would be unethical to withhold treat-ment from a woman with known diagnosed infection and it is impossible to know when an infection was acquired when de-tected by screening, evaluations of the natural history of chla-mydia will be difficult. However, studies that provide insight on some aspects of natural history may still be possible [34, 35]. Prospective studies that have obtained genital specimens for other reasons (eg, human papillomavirus natural history studies or vaccine trials and HIV prevention trials) should be explored as opportunities for better understanding chlamydia natural history, provided that study participants have given consent to test stored specimens and that chlamydia screening has been offered in accordance with standards of care. Addi-tional natural history data on repeat infections would also be useful [3], because these could help assess the importance of rescreening efforts. Ultimately, we would benefit from more-specific information about when treatment needs to be pro-vided to prevent sequelae and whether treatment of long-stand-ing prevalent infection has a tangible impact on complications. Better data on natural history would be helpful in modeling, shaping, and evaluating screening strategies in the context of a comprehensive chlamydia control program; such data would inform targeting of screening, the optimal frequency of screen-ing, and resource allocation for screening women versus man-aging sex partners.
A critical component of research addressing chlamydia nat-ural history and the impact of chlamydia screening is our ability to accurately measure the sequelae of C. trachomatis infection. We desperately need better, noninvasive tools to measure the complications of chlamydial infection. Diagnosis of acute PID is notoriously subjective, insensitive, and nonspecific [23, 36]. Infertility has multiple causes and may not be recognized for years after a chlamydial infection has caused tubal damage, because the affected woman may not have tried to become pregnant. Ectopic pregnancy is a more clear-cut diagnosis than is PID and infertility and could be more easily used in linking with administrative inpatient and outpatient data to assess out-comes. However, ectopic pregnancy is an uncommon outcome and its timing also depends on efforts or behaviors associated
S166 • JID 2010:201 (Suppl 2) • Gottlieb et al
with becoming pregnant. Thus, ideally, we need tools not only to more accurately assess the sequelae observed as end-products (eg, PID, ectopic pregnancy, and infertility) but also to detect the intervening pathophysiologic processes leading to or pre-dictive of those sequelae. More proximal, noninvasive markers of tubal damage would be extremely valuable for natural history and screening intervention studies and for candidate vaccine trials [37]. In addition, having the ability to predict individuals at increased risk of sequelae or reinfection could lead to tar-geting strategies that identify those who need, for example, more frequent screening or more intensive follow-up of sex partners. Thus, a better understanding of the immunologic, host, and organism factors underlying pathogenesis and se-quelae and a search for relevant clinical markers could ulti-mately help guide targeted screening and control efforts [38, 39].
Acknowledgments
We thank Christy Cechman, for her assistance with the literature search, and Tom Peterman, for helpful discussion and critical review of the manuscript.
References
1. Centers for Disease Control and Prevention. Infertility Prevention Pro-gram, USA. http://www.cdc.gov/std/infertility/ipp.htm. Accessed 29 January 2009.
2. Low N; SCREen project team. Publication of report on chlamydia control activities in Europe. Euro Surveill 2008; 13(28).pii:18924; er-ratum: Euro Surveill 2008; 13(34).pii:18960,
3. Haggerty CL, Gottlieb SL, Taylor BD, Low N, Xu F, Ness RB. Risk of sequelae after Chlamydia trachomatis genital infection in women. J Infect Dis 2010; 201(suppl 2):S134–S155 (in this supplement). 4. Stamm WE. Chlamydia trachomatis infections in the adult. In: Holmes
KK, Sparling PF, Stamm WE, et al., eds. Sexually transmitted diseases. New York: McGraw Hill Medical, 2008:575–594.
5. Centers for Disease Control and Prevention. Sexually transmitted dis-eases treatment guidelines. MMWR Morb Mortal Wkly Rep 2006; 55. 6. Geisler WM. Duration of untreated, uncomplicated genital Chlamydia
trachomatis infection and factors associated with chlamydia resolution:
a review of human studies. J Infect Dis 2010; 201(suppl 2):S104–S113 (in this supplement).
7. Molano M, Meijer CJ, Weiderpass E, et al. The natural course of
Chla-mydia trachomatis infection in asymptomatic Colombian women: a
5-year follow-up study. J Infect Dis 2005; 191(6):907–916.
8. Morre´ SA, van den Brule AJ, Rozendaal L, et al. The natural course of asymptomatic Chlamydia trachomatis infections: 45% clearance and no development of clinical PID after one-year follow-up. Int J STD AIDS 2002; 13(suppl 2):12–18.
9. Centers for Disease Control and Prevention. Chlamydia screening among sexually active young female enrollees of health plans—United States, 2000–2007. MMWR Morb Mortal Wkly Rep 2009; 58(14): 362–365.
10. Low N, Bender N, Nartey L, Shang A, Stephenson JM. Effectiveness of chlamydia screening: systematic review. Int J Epidemiol 2009; 38(2): 435–448.
11. Østergaard L, Andersen B, Moller JK, Olesen F. Home sampling versus conventional swab sampling for screening of Chlamydia trachomatis
in women: a cluster-randomized 1-year follow-up study. Clin Infect Dis 2000; 31(4):951–957.
12. Scholes D, Stergachis A, Heidrich FE, Andrilla H, Holmes KK, Stamm WE. Prevention of pelvic inflammatory disease by screening for cervical chlamydial infection. N Engl J Med 1996; 334(21):1362–1366. 13. Clark KL, Howell MR, Li Y, et al. Hospitalization rates in female US
Army recruits associated with a screening program for Chlamydia
tra-chomatis. Sex Transm Dis 2002; 29(1):1–5.
14. Andersen B, Østergaard L, Puho E, Skriver MV, Schonheyder HC. Ectopic pregnancies and reproductive capacity after Chlamydia
tra-chomatis positive and negative test results: a historical follow-up study.
Sex Transm Dis 2005; 32(6):377–381.
15. Bakken IJ, Skjeldestad FE, Lydersen S, Nordbo SA. Births and ectopic pregnancies in a large cohort of women tested for Chlamydia
tra-chomatis. Sex Transm Dis 2007; 34(10):739–743.
16. Hillis SD, Owens LM, Marchbanks PA, Amsterdam LF, Mac Kenzie WR. Recurrent chlamydial infections increase the risks of hospitali-zation for ectopic pregnancy and pelvic inflammatory disease. Am J Obstet Gynecol 1997; 176:103–107.
17. Low N, Egger M, Sterne JA, et al. Incidence of severe reproductive tract complications associated with diagnosed genital chlamydial in-fection: the Uppsala Women’s Cohort Study. Sex Transm Infect
2006; 82(3):212–218.
18. Hillis SD, Nakashima A, Amsterdam L, et al. The impact of a com-prehensive chlamydia prevention program in Wisconsin. Fam Plann Perspect 1995; 27(3):108–111.
19. Kamwendo F, Forslin L, Bodin L, Danielsson D. Decreasing incidences of gonorrhea- and chlamydia-associated acute pelvic inflammatory dis-ease: a 25-year study from an urban area of central Sweden. Sex Transm Dis 1996; 23(5):384–391.
20. Chen MY, Pan Y, Britt H, Donovan B. Trends in clinical encounters for pelvic inflammatory disease and epididymitis in a national sample of Australian general practices. Int J STD AIDS 2006; 17(6):384–386. 21. Egger M, Low N, Smith GD, Lindblom B, Herrmann B. Screening for chlamydial infections and the risk of ectopic pregnancy in a county in Sweden: ecological analysis. BMJ 1998; 316(7147):1776–1780. 22. Meyers DS, Halvorson H, Luckhaupt S. Screening for chlamydial
in-fection: an evidence update for the US Preventive Services Task Force. Ann Intern Med 2007; 147(2):135–142.
23. Paavonen J, Westro¨m L, Eschenbach DA. Pelvic inflammatory disease. In: Holmes KK, Sparling PF, Stamm WE, et al., eds. Sexually trans-mitted diseases. New York: McGraw Hill Medical, 2008:1017–1050. 24. Sutton MY, Sternberg M, Zaidi A, St Louis ME, Markowitz LE. Trends
in pelvic inflammatory disease hospital discharges and ambulatory vis-its, United States, 1985–2001. Sex Transm Dis 2005; 32(12):778–784. 25. Burstein GR, Zenilman JM, Gaydos CA, et al. Predictors of repeat
Chlamydia trachomatis infections diagnosed by DNA amplification
test-ing among inner city females. Sex Transm Infect 2001; 77(1):26–32. 26. Niccolai LM, Hochberg AL, Ethier KA, Lewis JB, Ickovics JR. Burden
of recurrent Chlamydia trachomatis infections in young women: further uncovering the “hidden epidemic.” Arch Pediatr Adolesc Med 2007; 161(3):246–251.
27. Brunham RC, Rekart ML. Considerations on Chlamydia trachomatis disease expression. FEMS Immunol Med Microbiol 2009; 55(2): 162–166.
28. Peterman TA, Gottlieb SL, Berman SM. Commentary: Chlamydia
tra-chomatis screening: what are we trying to do? Int J Epidemiol 2009;
38(2):449–451.
29. Hu D, Hook EW, III, Goldie SJ. The impact of natural history param-eters on the cost-effectiveness of Chlamydia trachomatis screening strat-egies. Sex Transm Dis 2006; 33(7):428–436.
30. van Valkengoed I, Morre´ SA, van den Brule AJ, Meijer CJ, Bouter LM, Boeke AJ. Overestimation of complication rates in evaluations of
Chla-mydia trachomatis screening programmes–implications for
cost-effec-tiveness analyses. Int J Epidemiol 2004; 33(2):416–425.
31. Oakeshott P, Kerry S, Atherton H, et al. Community-based trial of screening for Chlamydia trachomatis to prevent pelvic inflammatory
